A new interpretation of dragonfly wing venation based upon Early Upper fossils from Argentina (Insecta: Odonatoidea) and basic character states in pterygote wings

EDGARF. RIEK 19 DufJL Street, Ainslie, Conberrcl, A .C.T.,2602, Austrcllia

AND JARMILAKUKALOVA-PECK' Department of Geology, Carleton Universih, Ottc~wcl,Ont.. Cnnndo K IS 5B6 Received May 4. 1983

RIEK,E. F., and J. KUKALOVA-PECK.1984. A new interpretation of dragontly wing venation based upon Early Upper Carboniferous fossils from Argentina (Insecta: Odonatoidea) and basic character states in pterygote wings. Can. J. Zool. 62: 1150- 1166. The oldest known odonatoid wings are described from the Namurian of Argentina: Eugeropteron lunatum Riek n.g. et sp. and Geropteron arcuatum Riek n.g. and sp. (: Meganeurina: Eugeropteridae n, fam.). The wings are gener- alized and support a reinterpretation of the venation of living as being fully homologous to that of other pterygotes and closely related to Ephemeropteroidea, but different from . Therefore, Paleoptera is a valid phylogenetic unit, and Odonatoidea and Ephemeropteroidea are sister groups.

RIEK.E. F.. et J. KUKALOVA-PECK.1984. A new interpretation of dragonfly wing venation based upon Early Upper Carboniferous fossils from Argentina (Insecta: Odonatoidea) and basic character states in pterygote wings. Can. J. Zool. 62: 1 150- 1 166. On trouvera ici la description des plus vieilles ailes odonatoi'des jamais rencontrees: il s'agit d'ailes d'Eugeropteron lunatum n. gen. et n. sp. et de Geropteron arcuatum n. gen. et n. sp. Riek (Meganisoptera: Meganeurina: Eugeropteridae n. fam.) provenant du Namurien d'Argentine. Les ailes suivent le plan de base et donnent lieu a une reinterpretation de la nervation des Odonates rkcentes selon laquelle les ailes des Odonates recentes seraient entierement homologues des ailes d'autres ptkrygotes, trks rapprochees des ailes d'Ephemeropteroidea, mais differentes des ailes de Neoptera. L'appellation Paleoptera coiffe donc un ensemble phylogenktique valide et il est juste de considerer comme des groupes-soeurs les Odonatoidea et les Ephemeropteroidea. [Traduit par le journal]

Introduction Pterygote wing articulation was recently studied by Most entomologists follow Martynov (1924) and agree Kukalova-Peck ( 1983), who compared modern and Paleozoic that the pterygotes are naturally divided into Paleoptera and dragonflies with modern and Paleozoic Ephemeroptera, Paleo- Neoptera (Hennig 198 1 ). Nevertheless, some consider Paleo- zoic Paleodictyopteroidea (Paleodictyoptera, , ptera not to be a natural phylogenetic group, but rather a grade , and Permothemistida (=~rchodonata~)), (Lemche 1940; Boudreaux 1979; Rohdendorf and Rasnitsyn and primitive modern Neoptera. Multiple comparisons of artic- 1980; Rasnitsyn 1980; Kristensen 198 1 ; Matsuda 1970, 198 1 ). ular systems in different evolutionary stages showed that the Sometimes Odonata are assumed to be closer to Neoptera than odonatoid articulation evolved in a way similar to that of other to the Ephemeroptera (Kristensen 198 I), while other times, the Paleoptera, but unlike that of Neoptera. wings and wing articulation of Odonata are judged to be so The position of odonatoids within the Paleoptera, namely different that wings are believed to have originated whether they are more closely related to the paleodicty- twice: in the pterygote ancestor and in the odonatan ancestor opteroids or to the ephemeropteroids, is not clearly shown in (La Greca 1980; Matsuda 198 1 ). the articulation (Kukalova-Peck 1983). However, the closer These last two unorthodox evolutionary ideas find support in relationship to ephemeroids is evident in the veinal characters, the erroneous belief that the odonate wing venation and wing if the odonatoid veinal system presently used is corrected ac- articulation are unique and cannot be fully homologized with cording to the clues found in the fossil record. the other pterygote orders. The goal of this paper is to provide documentation that the We contend that wings and wing articulation originated at odonatoid venational system does not lack any veins as is the same time and brought winged into existence as a presently believed, and that it shares the ground plan with the group; therefore, they are considered as the basic apomorphic ephemeroids . characters of the pterygotes. It is believed that the split into Paleoptera, Neoptera, and the major ordinal evolutionary lin- Odonatan wing venation eages was initiated mainly by the radiation of the pterygote The interpretation of odonatan wing venation has a long proto-wings into diverse flight-adaptive structures (Martynov history of disagreement. Comstock and Needham ( 1898, 1899) 1924; Kukalova-Peck 1978, 1983; Hennig 198 1 ). If these con- erroneously assumed that veins are formed around tracheae and clusions are correct and wings are monophyletic, then odonatan wing venation and articulation must be fully homologous to 'we follow Sinichenkova (1980) in calling the order Permo- that of the other pterygotes. themistida Martynov, 1938, rather than Archodonata Martynov. 1932, because the group belongs in the paleodictyopteroids and has 'Author to whom reprint requests should be addressed. no relationship to Odonata. RIEK AND KUKALOVA-PECK 1151 that the nymphal tracheation recapitulates the ancestral vena- of with R, and the radial vein as the plesiomorphic double tion of adults. In fact, the preveinal blood channels in juveniles radius of primitive Paleodictyoptera (Figs. 6, 7) and fossil are invisible but present, and tracheae penetrate into them and living Ephemeroptera, i.e., composed of two radial stems secondarily and sometimes quite erratically. An erroneous RA and RP, the odonatan venation will lose its "strange" char- tracheae-based veinal system was adopted for odonates by acter and will become related to Ephemeroptera and Paleo- Needham ( 1903). Owing to vagaries of tracheation, the wings . Both "missing" veins MP- and CuA+ will then were interpreted as being completely unrelated to those of other be present in the venation, and the scheme will be at last fully pterygotes, i.e., RP (Rs)' was a convex vein crossed by M, and homologous with that in the other pterygote orders. M2, etc. Later on, numerous attempts were made by a number The original, generalized odonatan veinal scheme is indi- of authors to provide a homologous veinal scheme, namely cated in many Paleozoic Meganisoptera, but especially in- by Lameere (1922), Tillyard (1914, 1915, 1916, 1917, 19220, formative are two highly generalized (and probably also 19226, 1925a, 19256, 1926, 1928, 1935), Tillyard and the oldest known) dragonfly wings, which were recently Fraser ( 1938 - 1940), Martynov ( 1924), Carpenter ( 1 93 1 , found in the Early Upper Carboniferous deposits of Argentina. 1938, 1939, 1943n, 19436, 1944, 1947, 1960), Forbes (1943), Their venation represents the "missing link" between the Zalessky ( 1944), Fraser ( 1938, 1948, 1957), Needham ( 1903, typical odonatoid pattern and that of the ephemero-odonatoid 1951), Hamilton (1971, 1972), and Carle (1982). The inter- common ancestor. The Argentinian fossil dragonflies are pretations differ considerably, but none of them has been found described in the following text and their venation is compared to be correct in comparison with the venation of the oldest and with fossil and modern odonates, ephemeropteroids, and most generalized fossil dragonflies, which are described in this paleodictyopteroids. paper. One reason why there are so many seemingly feasible interpretations of odonate venation lies in the fact that in mod- Occurrence of the fossils ern dragonflies the primary veins alternate with the secondary Two very well preserved and almost complete insect wings intercalar veins and an occasional veinal supplement, and can- were found near Malanzan, Cuesta de la Herradura, La Rioja, not be readily distinguished from each other. However, sec- Argentina, in the carbonaceous group of the La Divisoria ondary veins are not fully tracheated and are sometimes miss- Member of the Malanzan Formation, by Drs. S. Archangelsky ing in the more primitive fossil dragonflies. Another reason is and R. Leguizamon. The insects were in association with a the basal parallel alignment of the fused veinal stems, which in plant assemblage including Vqjnowsba argentina Archangel- modern dragonflies is accompanied by disappearance of the sky & Leguizamon , Ginkgophvllum diuzii Archangelsky & original veinal crossings between these stems. If the crossings Arrondo, and Botrychiopsis u)eissiana Kurtz, as well as a were present, they would indicate the pattern under which the number of as yet undescribed plant species (S. Archangelsky, original fusions took place. The contribution of the fossil personal communication). dragonflies introduced here is that they (i) preserve ancestral Previous records of this fossiliferous horizon mentioned a character states of venation devoid of most secondary veins, or Early Carboniferous age, whereas Archangelsky and (ii) show the original pattern of veinal fusions before the and Leguizamon ( 197 I) thought that the strata might be dated crossings disappeared. Thus, generalized fossil dragonflies near the Westphalian -Stephanian boundary. Azcuy ( 1975) give clues to how the venation evolved. referred the horizon to the Namurian (Early Upper Carbon- Presently, the most widely followed odonatan veinal scheme iferous) after a palynological analysis. is that of Till yard and Fraser ( 1938 - 1940). This is based upon Both fossil wings are very similar to the wing of Erasipteron the veinal pattern of rather specialized Paleozoic dragonflies larischi Pruvost, 1933: Meganisoptera ( =Protodonata) from and shows the venation as being almost homologous to other the Lower Westphalian A (Bashkirian) of Czechoslovakia, pre- pterygotes except for two main veins: the media posterior MP- viously known as the oldest dragonfly, but they are more gen- and cubitus anterior CuA+, which are interpreted as missing. eralized. It seems, therefore, that the horizon may be older than Beside this nonhomologous feature, the Tillyard-Fraser veinal early Westphalian, and may be at least of Namurian age, as system does not display any relationship between odonates and shown by the palynological analysis. From a study of the in- other pterygote orders. Conspicuously, there is no similarity sects alone, an even older age is not excluded. Presently, there between Odonata and the other surviving order of Paleoptera, are only about a dozen winged insects known with certainty to the Ephemeroptera. Because of this theoretically insurmount- occur in Namurian strata. Thus, the primitive meganisopterids able difference in an important set of apomorphic characters, from Argentina are among the earliest known pterygotes, if not some entomologists felt that the odonatoids cannot be classified actually the earliest ones now recorded in fossil collections. with Paleoptera or even with (La Greca 1980; Matsuda 198 1). In contrast, theoretical cladists like Hennig Systematics (1981) assumed that dragonflies and are related, the Odonatoids form a very distinctive superorder including differences in nomenclature of venation notwithstanding. the diverse, mostly Paleozoic stem order Meganisoptera During the work with Upper Paleozoic and odon- (=Protodonata), and the less diverse daughter order Odonata, atoid wings, one of us (Riek) became convinced that two veins which has survived until modern times. New, rich material of in Tillyard-Fraser veinal system, the media and cubitus, were Meganisoptera and Mesozoic Odonata has been recently re- misunderstood (Riek 1970, p. 173, Fig. 8.2). If the medial searched by L. N. Pritykina (Paleontological Institute, Acade- stem is interpreted as in Paleozoic Ephemeroptera (Figs. 10, my of Sciences USSR, Moscow), but the results are only partly 20), i.e., as fused near the base with the cubital stem instead published (Pritykina cited in Rohdendorf and Rasnitsyn 1980; Pritykina 1981). The systematics of fossil odonatans was pre- 3~nemended veinal nomenclature, consistently using suffix A for viously based almost entirely on wing venation (Carpenter anterior veinal sectors (i.e.. RA), and suffix P for posterior sectors 1939, 1943u, 19436, 1954; Rohdendorf 1962), but in fossil (i.e., RP), is used in this paper (see Kukalovli-Peck 1983). collections from France and Russia, several specimens are pre- 1152 CAN. J. ZOOL. VOL. 62. 1984 served with their bodies. The newly added characters, quite The oldest known dragonflies from the Namurian of Argen- understandably, throw a different light on the systematics of tina, presented here (Figs. 1 - 3, 13 - 1 c)), are members of the Meganisoptera, as follows. most primitive meganisopterid suborder Meganeurina and According to Kukalova-Peck ( 1983), large Meganisoptera are closely related to Erusipterorz larisc-hi Pruvost , 1933 are prognathous, and their prothoracic coxae are shifted far (Erasipteridae Carpenter, 1939) from the Westphalian A anteriorly under the head so that their long, strong, and spiny (Bashkirian) of Czechoslovakia (Fig. 4) and to Erusipterella forelegs can reach forward (a paper documenting this is in piesbergensis Brauckmann, 1983 from the Westphalian of preparation). In contrast, Odonata are hypognathous and have West Germany. Kukalova ( 1964, Fig. 2), found remnants of an relatively weak and much shorter legs, which serve as a archedictyon in the wing membrane. "fishing" basket. Pritykina ( 198 1 ; and personal communication with Kukalovi-Peck in Moscow, 198 1) discovered that Mega- ORDER Meganisoptera (=Protodonata) nisoptera had a deeply incised "unclosed thoracic pleural SUBORDER Meganeurina Handlirsch, 1906 sulcus" with an additional inflexion running parallel to it ante- riorly, while Odonata have a "closed pleural sulcus." The Meganeurina is the most primitive odonatoid suborder "double" pleural sulcus, which occurs in meganeurinid, pro- containing medium-sized, large, and the largest known odon- tanisopterid, protozygopterid, archizygopterid, and triadophle- atoids with a wing span of 7 1 cm (in Megcrrzeuropsis permiuna biinid ancestral dragonflies probably as a synapomorphy, is a Carpenter, 1939, from the Lower of Kansas). Nodus, unique feature in Pterygota and is a derived character. The arculus, and costal triangle are missing, but a pterostigma may "closed pleural sulcus" of living dragonflies developed perhaps be indicated. In the most primitive new family Eugeropteridae by further invagination of the original pleural sulcus and of and in related Erasipteridae, the venation carries relatively few the parallel inflexion. This feature may well be the basic branches and very few intercalar veins, and the membrane apomorphic character of the order Odonata as a daughter shows remnants of an archedictyon between the crossveins group. (Figs. 1 -3, 4, 13). In the derived families with secondarily The wings of Meganisoptera (Figs. 1-5, 11) lack an arculus enlarged wings, namely in Meganeuridae, Paralogidae, and and often lack a pterostigma, and the nodus is either completely Triadotypidae, the venation is enriched by many additional absent or weakly indicated; venation is generally more prim- intercalar branches and crossveins, which evolved secondarily itive: the veins forming the costal margin (PC, CA, CP, and from the membrane (Kukalovh-Peck 1983) (Fig. 5). In mega- ScA) are basally separated from each other into PC and CA, neuromorph families the section of CUP between M and Cu CP, and ScA, and do not fuse into the basal costal triangle; the stem and anal vein (CUPcrossing) is usually well defined, long, veinal stems are basally aligned in parallel more or less imper- oblique, and distinctly concave (Fig. 1 1 ). The free, basal fectly, and include a free basal portion of convex medial stem portion of the medial stem (M+) is probably always present (M+); CUP- is often recognizable between CuA+ and A (Figs. 1-5, 11). branches; and the anal brace is often more or less concavely Oc~currence:Tropical and probably temperate zone, Namu- bent. rian to Triassic. The wings of Odonata (Fig. 12A) have an arculus and a Families included: Eugeropteridae, Riek n. fam. ; Er- pterostigma and their nodus is pronounced. The costal margin asipteridae Carpenter, 1939; Meganeuridae Handlirsch, 1906; is basally fused into the basal costal triangle (Fig. 12B); the Paralogidae Handlirsch, 1906; Triadotypidae Grauvogel & veinal stems are basally well aligned in parallel; the basal Laurentiaux, 1952. portion of M is completely fused with Cu; CUP is not recog- nizable among the cubito-anal branches; the anal brace is Eugeropteridae, Riek n. fam. aligned in parallel with other veinal stems basally; and Type genus: Eugeropteron, Riek n. gen. secondary veinal supplements may occur in RP and MA-MP Diagnosis: Eugeropteridae differ from the closeIy related areas. family Erasipteridae in the veinal stems RA and RP running

FIG. I. A generalized odonatoid wing showing the basic apomorphic character of odonatoid venation: the zig-zag kink in CUP,fused to AAI and connected by a crossvein (cr) to CuA. The odonatoid foci of reinforcement were marked by veinal arches and struts: MA and RP, and CuA and M. Veins of the costal margin (PC and CA+, CP-. ScA+) were separated basally; RA and RP were separated basally, not adjacent in a double radius; the anal brace was veinal, composed of AA, AA' and AAI; AP was branched and concave. Note the similarity to Carboniferous Ephemeroptera (Fig. 20) in all veinal features except the kink in CUP. Meganisoptera: Meganeuromorpha: Eugeropteridae n. farn.; Eugeroprerum lunarum Riek n. gen. et sp. Holotype, hind wing fragment, length 35 mm, ?Lower Namurian of Argentina. Original drawing by E. F. Riek. FIG.2. Eugeropteron lunarum Riek n. gen. et sp. Holotype, fore wing fragment, length 13.5 mm. '?Lower Namurian of Argentina. Original drawing by E. F. Riek. FIG.3. A slightly more advanced odonatoid wing showing the zig-zag kink in CUPfused with the anal brace and also directly fused with CuA. This double fusion occurs in the wings of all recent Odonata. MA-, is replaced by an intercalar vein. Meganisoptera: Meganeuromorpha: Eugeropteridae n. fam. : Geropreron arc-uarum Riek n. gen. et sp. Holotype, hind wing fragment, length 45 mm. ?Lower Namurian of Argentina. Original drawing by E. F. Riek. FIG.4. A still more advanced wing, in which veinal arches and struts are transformed by fusions: MA and RP, and M and CuA. CUP- and CuA+ crossings are both discernible but the MA crossing can hardly be distinguished from a crossvein. The wing is typically odonatoid in appearance. Note the similarity to Permian Ephemeroptera (Fig. 10) in all veinal fusions. Meganisoptera: Meganeuromorpha: Erasipteridae: Erasipreron larischi Pruvost, 1933, Westphalian A of Czechoslovakia. Holotype, fore wing fragment length 55 mm, total length ca. 64 mm. Figure after Kukalova 1964, with veinal nomenclature changed. FIG. 5. A still more advanced wing of a large meganeurid dragonfly, diagramatically showing how the veins have fused. The MA crossing became fully aligned with the series of crossveins, the CUPcrossing stayed deeply concave and well distinguishable, and the CuA crossing was no longer recognizable. The secondary branches, intercalar veins, and crossveins served to fill out the enlarged wings. Note the absence of the basal costal triangle, nodus, subnodus, and arculus, and the presence of a long, well-defined CUP.The composite anterior axillary plate shows deep grooves between articular sclerites. Meganisoptera: Meganeuromorpha: scheme based upon specimen No. I/ 1983, Lower Permian of Obora, Czech- oslovakia. Collection of and original drawing by J. Kukalova-Peck. Sc A + ScP - 1154 CAN. 1. ZOOL. VOL. 62, 1984 close but not adjacent to each other; stem of M closely ap- foldable wings adapted for gliding. The wings show a relatively proaching Cu but not fused with any part of Cu; MA arched derived venation with intercalated veins and sclerotized anal towards RP but not fused with RP. brace. Relationships: The odonatoid character of the new family It should be noted that venational fusions RP and MA or M, Eugeropteridae is documented in the articulation by the frag- CuA and M (or Cu and M), and CUPand A (but not R and M ments of two axillary plates of primitive odonatoid type in the at the wing base), occur repeatedly in all fossil paleopterous holotype of Eugeropteron lunatum (compare Figs. 1, 13, 4 orders as well as in modern Paleoptera. Beside Odonata, they with 5 and I I), and in the venation by the presence of the are present in advanced Paleodictyoptera (i.e., in some Calver- typical odonatoid veinal anal brace4 formed by AA and fused tiellidae; Kukalova-Peck and Peck 1976, Fig. 2), Diaphanop- with a zig-zag kink in CUP, which in turn is fused with CuA terodea (i .e., in Elmoidae; Kukalova-Peck 1974), Megasecop- (Figs. 1-3). A veinal anal brace of this type is limited to tera (i .e., in Alectoneuridae; Kukalova-Peck 1975, Fig. 1 2), odonatoids, is found in venation of all representatives, and is, Permothemistida (i .e., Permothemistidae; Sinichenkova 1980, therefore, the basic apomorphic character of the superorder Fig. I), and in Ephemeroptera (Fig. 10). Likewise, intercalated (Figs. 1-5, 1 1, 12). Early ephemeroids (Fig. 10) have a veins occur not only in odonatoids and ephemeropteroids, but homologous anal brace formed by identical parts of anal also in specialized Paleodictyoptera, Megasecoptera, and in branches, but instead of a kink in CUPthere is a bulla. Very Permothemistida (Figs. 8, 9) (Kukalova-Peck 1978, 1983). probably, the common ancestors of odonatoids and ephemerop- The generalized venation found in Eugeropteridae provides teroids were gliding paleopterans, in which the meeting point a morphological missing link between the venation of Odon- of the anal brace, CUP, and CuA became very important for atoidea and that of Paleozoic Ephemeroptera (Figs. 1 , 3, 4, 10, flight. When the structures responding to flight mechanics dif- 11, 12, 13, 20) and of Paleodictyopteroidea. These characters ferentiated, ephemeropteroids retained a bulla in CUP, and include (i) plesiomorphic pterygote characters, such as the odonatoids formed a kink in CUP. veins composing the costal margin (PC, CA, CP, and ScA) In contrast to this, all paleodictyopteroids have a primitive, separated basally from each other, RA and RP stems separated nonveinal, sclerotized anal brace, which is merely a thickening basally from each other, and ScP- fused with C distally with- of the membrane in the basal corner of the anal area. Prim- out a nodus; (ii) characters acquired mostly by parallel evo- itively, the anal branches crossing the sclerotization are visible lution, typical for Paleoptera, such as MA arching towards RP, (Figs. 6, 7), but in more advanced paleodictyopteroids they CuA arching towards M, and very pronounced, regular cor- might be obliterated. The anal sclerotized brace is often poste- rugation; and (iii) ephemero-odonatoid synapomorphic charac- riorly delimited by a convex bar (Figs. 6, 7, 9: S), which is ter, the veinal anal brace AA meeting CUPat a flight-important variably shaped, straight, curved, or V-shaped (Kukalova-Peck area. Eugeropteridae differ from all ephemeropteroids in the 1 969, 1970a, 19706). The bar runs transversely across a vary- presence of the odonatoid basic apomorphic character, a ing number of anal branches, usually at the generic or specific zig-zag kink in CUP fused proximally with the veinal anal levels. brace AA and distally with CuA, which is typical only for The new family Eugeropteridae displays a venational pattern odonatoids. very similar to the related family Erasipteridae (Fig. 4) in the number of branches and type of crossveins. It differs in the Eugeropteron Riek n. gen. primitive absence of veinal fusions: the RA and RP stems are Type species: Eugeropteron lunatum n . sp. basally close to each other but separated, not yet adjacent as in Diagnosis: Hind wing: Sc ending on costal margin beyond more advanced families; MA arches towards RP and is con- midwing; stems of RA+ and RP- widely separated from each nected with it by two struts, but fusion has not yet occurred; other basally, approaching each other at about one-sixth of the CuA arches towards M and is connected with it by a short strut, wing length; stem of M+ straight, aligned in parallel with RP; but no fusion has yet occurred. MA+ and MP- diverging from each other shortly before the Except for the presence of an odonatoid basic apomorphic point of divergence of RA+ from RP-; MA+ arched towards character, the kink in CUP, which is fused between the anal RP- and connected with it by two struts; MA+ with three brace and CuA, the veinal pattern of Eugeropteridae is gener- branches; Cu+ stem concavely bent; CuA+ and CUP- div- alized (Figs. 1 - 3). In addition to the Ephemeroptera, it is also erging from each other close to base; CuA+ arched towards quite similar to that of the largest and most primitive Paleo- M+ and connected with it by a short strut; CuA+ and CuP- dictyoptera, the Homoiopteridae (Fig. 6), and to Permo- crossing cubito-anal area towards AA; CUP after fusing with themistida (Figs. 8, 9). The Paleozoic order Permothemistida AA turns immediately distally and then posteriorly in a zig-zag (=Archodonata) is a relatively small paleopterous order with kink, and continues posteriorly as a simple vein; CuA con- haustellate, paleodictyopteroid type of mouthparts, and non- nected with zig-zag kink in CUPby a crossvein, sending off two branches; AA1+ sending off AA, fused with CUP,and simple 4~heterm "brace" defines a supporting, usually transverse struc- AA,; AA"+ divided into forked AA3 and simple AA,; AP- ture, which evolved in the flying wing in response to aerodynamic starting from anal basivenale, with a short terminal fork; JA+' forces. Braces can be formed by crossveins, veinal arches and struts, with a terminal fork; JP- simple; "first antenodal" crossvein and by veinal crossings and fusions, or, can be formed by a sclero- running from costal margin to R and continued to RP; cross- tization arising from the membrane. Bracing is opportunistic. Mem- bers of the same taxon usually have braces in similar position, but they veins simple and openly spaced; intercalated veins few, includ- can be formed by different elements, such as an arch and a crossvein, ing about 3-4 veins between RP branches, one between CuA or by a direct veinal fusion. Braces are often composite and built from several elements, like sections of branches combined with crossveins. "he presence of an all-pterygote, homologous jugal area with They are not always rigid, but may be cut by flexion lines, bullae, etc. branching jugal veins was recently discovered in Paleoptera by The pattern of braces coincides with the pattern of foci of reinforce- Kukalova-Peck (1983). Therefore, a jugal area is plesiomorphic for ment needed to make the wing flightworthy. the pterygotes. RlEK AND KUKALOVA-PECK 1155 branches, and one distally from AA2. Archedictyon reduced Divisoria Member, Malanzan, Cuesta de la Herradura, La and barely noticeable. Rioja, Argentina. Fore wing remnant: Kink in CUPflanked distally by a cross- vein as in the hind wing; AA2+ with sub-branches, Aj and A, Why are Eugeropteridae true odonatoids? simple; AP- simple; JA+ simple; crossveins fewer than in the In the of living insects, the whole is hind wing. almost always available for morphological and systematic Wing articulation: anterior axillary plate and posterior axil- study. In paleoentomology, the material is often limited to lary plate of meganisopterid type. wings, and many taxa are based solely upon wing venation. For Species: Eugeropteron lunatum, n. sp., ?Lower Namurian more than a century, paleoentomologists have devoted atten- of Argentina, probably temperate zone of the Pangean tion to the characters of venation and searched for criteria to supercontinent. separate "stable" from "variable" characters. It has also been known for a long time that wing venation between and within Eugeropteron lunatum Riek n. sp. orders is widely convergent. The development of venational Figs. 1, 2, 13, 14, 15, 16 convergence within separate orders was followed through time, Holotype: Specimen No. 12886, Museo La Plata, Fossil because the fossil record provided the unique opportunity to Invertebrate Division, La Plata, Argentina. Fragment of fore examine character changes as they evolved. This research has wing showing portion of cubital and anal area (Fig. 2), hind brought abundant evidence that the clues to wing venation of wing with a fragment of wing articulation of meganisopterid recent orders lay with their Paleozoic stem groups. type, missing apical part (Fig. 1). Recent orders are offshoots of a very limited number of Description: By monotypy, the same as that of the genus. Paleozoic insects and are usually not older than the Permian or Fore and hind wings almost homonomous, hind wing slightly Triassic. This also includes the order Odonata. As a rule, the broader; AA2 in the fore wing with 6 terminal branches distally stem groups show more varied venation (at higher systematic from the zig-zag kink in CUP. Wing margin as well as mem- levels), which is richer in branches. Instead of fused veins, this brane without obvious setae or spines. Hind wing fragment, venation frequently has unfused veins, arched veins, and struts, length 35 mm; width 13 mm; total length ca. 38 mm. or shows easily interpretable fusions. Occurrence: ?Lower Namurian; Malanzan Formation, La For understanding veinal evolution at ordinal levels, the Divisoria Member, Malanzan, Cuesta de la Herradura, La recognition of ordinal basic apomorphic characters is crucial. Rioja, Argentina. The theoretical presumption is that each order group originated by the development of at least one, but usually several, new Geropteron Riek n. gen. character states, by which the group is defined. A basic apo- Type species: Geropteron arcuatum, Riek n. sp. morphic character is distinguished from all other apomorphic Diagnosis: Hind wing: Wing venation differs from the characters by the fact that it occurred earlier and that it is shared closely related new genus Eugeropteron in longer ScP- by all members of the group, extinct as well as extant. It should touching RA+ before terminating on costal margin at about be noted that insect wings evolved from veined, movable proto- three quarters of the wing length; MA+ with two primary wings that were not flight adapted (Kukalova-Peck 1978, branches and one intercalar vein; zig-zag kink on CUP more 1983). Changing nonflying flaps into a flight organ required a pronounced so that CUPfirst fuses shortly with the first branch regrouping of the veins into a pattern that would be flightwor- of AA and then fuses shortly with CuA; CUP separates from thy. Therefore, wings of all pterygotes are a combination of CuA as a simple concave vein directed obliquely towards pos- symmetrical (nonadaptive) and asymmetrical (flight-adapted) terior margin; AAt+ sending off AA, fully fused with CUP, veinal elements (Kukalova-Peck 1983). and forked AA2, with one intercalar vein; AP concave. Hind In all flying insect wings, there is a certain pattern of areas wing relatively narrower in the anal area. or foci that have to be reinforced for flight. The reinforcements Relationships: Geropteron is more advanced than the closely are often provided by fusions among veins, which originated related genus Eugeropteron in having deeper zig-zag kink in partly at the very beginning of flight, but mostly in the process CUP- fused directly with CuA+ instead of being connected of diversification of orders. Since adaptation for flight was with CuA by means of a crossvein. Anal brace of Geropteron orchestrated by identical aerodynamic laws, veinal fusions is a morphological transition between that of Eugeropteron and within orders formed in parallel, with the result that they be- Erasipteron . came widely convergent. Because of this convergence, the Species included: Geropteron arcuatum Riek n. sp., ?Lower wings of many pterygote orders look alike to an untrained eye. Namurian of the Pangean supercontinent. Many times it was also observed that the older the wings are, the less distinguishable they become (Figs. 1, 2, 6- 1 1). The Geropteron arcuatum Riek n. sp. Figs. 3, 17, 18, 19 best example of an increasing similarity is the well-known Holotype: Specimen No. 12885, Museo La Plata, Fossil Paleozoic wing-based "waste-basket" order , Invertebrate Division, La Plata, Argentina. Hind wing without which contains stem groups of plecopteroids, orthopteroids, proximal part of posterior margin. blattopteroids, hemipteroids, and neuropteroids, not because Description: By monotypy, the same as that of the genus. they had not been well separated from each other, but because Wing margin and membrane without obvious setae or spines. the wing venation is still very little differentiated. Eventually, There appears to be a bulla at the base of the first anterior even venational differences between Paleoptera and Neoptera branch of RP. Archedictyon reduced, indicated weakly be- tend to disappear (Kukalova-Peck 1983). This process of tween the branches of AA and elsewhere. "vanishing characters," theoretically, should continue until the Dimensions: Hind wing length, 45 mm; width, 12 mm; total ultimate ancestor would have only the basic (definitive) apo- length, ca. 46 mm; total width, ca. 13 mm. morphic characters plus plesiomorphic (ancestral) characters. Occurrence: ?Lower Namurian, Malanzan Formation, La This "reversed evolutionary process" as documented by fossils 1156 CAN J Z001. VOL. 62. IOX1 is, of course, in accord with the evolutionary basis of system- As expected from theoretical considerations, the archetypal atics, and provides a definitive proof to the monophyly of odonatoid foci of wing reinforcement are identical to those insect wings. present in the ephemeropteroids. In addition, in Carboniferous Recent entomologists, who have the luxury of examining the Ephemeroptcra (the most primitive known, Fig. 20), the wings whole organism, sometimes tend to mistrust the capacity of show a strikingly parallel pattern of arches and struts as in wing venation to provide reliable criteria for taxonomic work. Eugeropteridae (Figs. 1 - 3), while Permian Ephemeroptera This is true in some insect groups (i.e., the fore wings of (Fig. 10) have them replaced by fusions as in Erasipteridae have very variable venation). but in other groups (Fig. 5). Note that "double radius" is very pronounced in the venation is like a thumbprint and provides reliable charac- Carboniferous maytlies, less visible in Permian forms, but still ters from ordinal down to specific levels. It is a well-known recognizable in well-sclerotized modern forms. Also note that fact that Odonatoidea, Ephen~eropteroidea,and Paleodictyop- the stem of M starts in maytlies well posteriorly from the teroidea have a very characteristic wing venation and that they "double radius," and therefore the latter cannot be interpreted can often be identified even when the body parts are not exam- as R and M as it is in some contemporary publications. ined or not available. A trained entomologist can readily recog- Ephenlel-o-odonatoids share a basic synapomorphy, which is nize a wing of any recent or Mesozoic dragonfly based on the the veinal anal brace AA meeting CUP at a flight-important wing veins. Late Upper Carboniferous and Permian mega- area. In contrast, all paleodictyopteroids have a sclerotized anal neurid odonatoids still "look" like dragonflies. even if they lack brace. From an evolutionary viewpoint, Eugeropteridae are an the nodus, subnodus, costal triangle, veinal supplen~ents.ar- excellent exanlple ol'a missing link and are probably very close culus, alignment of veinal stems, and have a well pronounced to a hypothetical early odonate. They seem to be generalized in CUP(compare Fig. 5 with Fig. 12). Because of this obviously all aspects except, perhaps, in the paucity of veinal branches, more primitive, but not very "strange" set of characters, a because other primitive Paleoptera as well as Neoptera have rigorous recent taxonomist might be tempted to regard the richly dichotomously branched veins. meganeurid veinal pattern as original for odonatoids. However, The venational pattern of Eugeropteridae provides an out- this presumption is not at all probable. Compared with the standing example 01' a succession of characters, theoretically wings of the most primitive Paleodictyoptera (Figs. 6, 7) forn~ulatedby Hennig (198 1 ). While the odonatoid nature is as well as Ephemeroptera (Figs. 12. 20) and Neoptera, the established by the presence of a basic apomorphic character, meganeurid wing venation is highly derived in many charac- the future apomorphic characters are only suggested and ters. Interpreting the origin of such a complex change by a become more pronounced later through a process of the megamutation, as may be favored by some authors. is highly autapomorphic accumulation of small character changes. unlikely. Autapornorphy in odonatoids is extremely pronounced and is at It is a well-known fact, based upon the comparative study of the root of all problems with homologization to other insects. many groups of Paleozoic insects, that the stem groups of The trrinsformation of veins from separated to arched and living orders may have unfused veins, or that the fusions are then to fused, and of veinal crossings from discernible to lost, sometimes preshadowed by veinal arches and struts. The is an evolutionary process observed in all Paleozic paleopterous Argentinian ancestral dragonflies (Eugeropteridae) occur in the orders and is demonstrably parallel. This fact indicates that Namurian and they may be the oldest known pterygotes. 'Their ancestral pterygotes diverged very early, probably already dur- wings show a combination of the odonatoid basic apomorphic ing the transition of proto-wings to flying wings. Major ptery- veinal character and the incipient pattern of odonatoid apo- gote lineages probably started with venation virtually devoid of morphic veinal fusions, represented by veinal arches and struts. fusions, except perhaps the veins reinforcing the anterior wing These are a kink in CUPfused between the veinal anal brace and margin. This consideration is important for systematic evalu- CuA, and the arched MA to RP and CuA to M. However, we ations of recent wings. because it implies that the veinal pat- know that arches and struts are interchangeable with veinal terns mostly evolved at lower than ordinal levels. To estimate fusions, because braces are opportunistic. This archetypal which way the adaptation was directed in a particular post- odonatoid pattern revealed a hidden character of basic impor- Paleozoic group. it was suggested that one compare a given tance, which could not have been deduced from recent and specimen first with the model of the proto-wing, then with the Mesozoic dragonfly venation: the kink in CUPbuilt between model summarizing wing adaptations during the Paleozoic, and AA and CuA and its fusion with both of these veins. This finally with the key representatives of the particular living discovery made homologization of odonatoid venation with order. In this way, the Paleozoic phase of evolution, which is that of all other pterygotes smooth and easy. so important for correct evaluation of character states, can be While additional fusion (a brace) between CUP and AA is incorporated into phylogenetic considerations and systematic present in all pterygotes except 1110st Paleodictyopteroidea, decisions (KukalovA-Peck 1983, Figs. 13 - 15). CuA and CUPseparate usually at or near the cubital basivenale, Further evidence that Eugeropteridae are true odonatoids, and then no longer brace with each other. However, the pres- even if they do not "look" like recent Odonata, is found in the ence of a bracing CuA and CUPin the Odonatoidea is by no remnants of the wing articulation attached to the wing base means unique. It occurs also in some primitive Paleozoic (Figs. 1, 13, 13). In contrast to other insects, this articulation Neoptera and is well developed especially in Geraridae (Pro- includes two axillary plates, the anterior and the posterior. torthoptera), recently revised by Burnham ( 1983, Figs. 17a, The posterior axillary plate is present also in other gliding 19, 25). In the large fore wings of this family, CUPis forked Paleoptera, but is absent from the Neoptera, which have axil- and CUP,-?fuses temporarily with CuAS and separates from lary sclerites of a different composition (Kukalova-Peck 1983). it again as a concave branch, simple or forked. While the The anterior axillary plate is a complex supersclerite formed by presence of a CuA and CUP brace is convergent among fusions of four dorsal articular sclerites with ventral articular pterygote orders, the bracing by means of a kink in CUP is sclerites, the basalaria (Kukalova-Peck 1983, Figs. 16B, unique and a hallmark of the Odonatoidea. 16E, 2 1 ). This plate is unique to odonatoids and, in all proba- bility, is the basic apomorphic character of the odonatoid artic- Paleodictyoptera, RA is always convex, while RP is either ulation. The four articular sclerites that compose the odonatoid concave or levelled, but never convex. There is no doubt that anterior axillary plate are movably hinged to each other in the concave position of the basal stem of RP is primitive. The ephemeropteroids, paleodictyopteroids, and in Neoptera. problem is how to explain the convexity of RP stem in While the anterior axillary plate of recent odonatoids shows Ercisipteron and all younger odonatoids. almost no sutures among the original sclerites, meganeurids Recently. Kukalova-Peck (1983) provided evidence that the have deeply incised and partly flexible grooves separating fluting of veinal stelns near the base may be variable at lower the four dorsal sclerites and the ventral basalaria (Figs. 5. 11) than ordinal levels, especially in Paleozoic insects. While ScP (detailed report in preparation by J. Kukalova-Peck). Eugerop- is always concave and R (or RA) always convex, the stem of teridae show the remnant of an odonatoid anterior axillary M can be convex, levelled, or concave, and very short or long; plate, which is, quite predictably, close to the meganeurid type the stem of Cu convex, concave, or missing; and the stem of (Figs. 1, 5, 11, 13, 14, 15). A very short and convex or missing. There is little doubt that changes in fluting, presence of stems, and length of stems are Fluting of basal veinal stems derived and are controlled by aerodynamic laws, but the cause It should be noted that under the generalized condition and effect are only superficially known. present in the Eugeropteridae, radial sectors are completely Seven pairs of wing veins have been found by now, one at separated and the entire posterior radius (RP) is concave, con- a time, to be completely separated basally into two sectors, trary to that in all other odonatoids, in which radial sectors are namely PC, C, Sc, R, Cu, A, and J (Kukalova-Peck 1983). The adjacent and RP is always convex. In the adjacent basal stems only exception is M, which always has at least a short basal of the radial sectors (double radius) of Ephemeroptera and stem. Very probably, it is only a question of time before M will

FIG.6. Primitive Paleodictyoptera with two radial sectoral stems RA+ and RP+ adjacent in a double radius. Venational foci of reinforcement were M and RP, and Cu and M, both placed distally from the wing base. The anal brace was a sclerotized basal corner of the anal area (dotted) delimited by a convex bar (S). Note the lack of a brace between AA and CUPand the presence of a well-defined jugal area with two jugal veins. Paleodictyoptera: Homoiopteride: Thesoneura americ.crner Carpenter. 1943. Westphalian C-D of Illinois. Holotype, hind wing, total length 1 10 mm. Original drawing from holotype, Kukalovi-Peck and Richardson 1983. FIG. 7. A paleodictyopterid wing documenting the transition between a double (primitive) and a single (derived) radial stem: RA+ was superimposed on KP- basally, but the sectoral stems were adjacent as a double radius more distally before they diverged. AP was preserved in the primitive concave position within the sclerotized anal brace, but became secondarily levelled (5)after crossing the convex bar (S). Paleodictyoptera: Homoiopteridae: Homoioptern giganteel Agnus, 1902, Stephanian of Commentry. France. Holotype, prothoracic and mesothoracic wing, total length of the latter 187 mm. After Kukalova-Peck and Richardson 1983. FIGS.8 and 9. Advanced paleodictyopteroid Permothemistida had an identical ground plan of veinal foci of reinforcement as did the ephemero-odonatoids (MA and RP, CuA and M. AA and CUP),but had paleodictyopteroid type of sclerotized anal brace (S). Note the highly specialized, intercalated vein IA (dotted) replacing part of AA and of AP, which is probably indicative of a unique style of flight. Fig. 8. Permothemistida: Diathemidae: Diclthernidin monstruoscl Sinichenkova. 1980; Lower Permian of Perm. U.S.S.R. Holotype, fore wing, total length 17. 2 mm. Fig. 9. Permothemistida: Permothemistidae: Permothemis libelluloide.s (Martynov. 1932). Upper Permian. SE of European U.S.S.R. Holotype, fore wing length 17.3 mm. Original drawings from holotypes by J. Kukalovli-Peck. FIG. 10. Permian mayflies had the anal brace AA and AA' and AAI fused with CUPat a bulla, and extended to CuA by a crossvein. The identical veinal anal brace, the veins of the costal margin separated basally, the double radius, and the pattern of veinal fusions MA and RP, and CuA and M, demonstrate the close relationship to early Odonatoidea. Ephemeroptera: Protereismatidae: Protereismcr sp.. Lower Permian of Kansas. Hind wing, total length ca. 20 mm. After Kukalova-Peck 1983. FIG. I I. The giant meganeurid dragonflies had a composite anterior axillary plate differently shaped than that of recent odonates, with deeply incised grooves between articular sclerites. Vcins composing the costal margin (PC and CA+, CP-, ScA+) were always separated basally. The double radius resembled that of the odonates, but the stems of M + and Cu- were basally separated, the MA crossing was rarely discernible and the CuA crossing was not discernible from crossveins, while the CUPcrossing was always concave and well pronounced. Note the primitively concave AP and presence of convex and concave jugal veins. Meganisoptera: Meganeuromorpha: Meganeuridae: Megatypus schucherti Tillyard. 1925, Lower Permian of Kansas. Holotype, hind wing, maximum width 35 mm. Original drawing from holotype by J. Kukalova-Peck.

FIG. 12A. A corrected interpretation of the venation of recent dragonflies. The veins of the costal margin PC and CA and CP and ScA are fused basally into a costal triangle: ScP forms a kink at the nodus (N) and fuses with the costal margin; RA is convex, simple; RP is basally secondarily convex and adjacent to RA, then shortly superimposed by MA+ and forming a kink RP and MA+. from which RP emerges as a concave sector; RP is dichotomously branched into RPI. RP2, and RP2- 4; M and Cu are fused basally into a concave stem M and Cu-, then CUPdiverges posteriorly as the CUPcrossing. then CuA diverges posteriorly but the crossing is lost, then MA diverges anteriorly towards RP and the crossing is lost; RP and MA+ form a kink, from which MA emerges as a convex, simple sector; MP runs posteriorly from the arculus, forms a kink and continues as a concave, simple sector towards posterior margin; the arculus is composed of a kink in RP and MA+, MA+, and of a crossvein; CUPfuses with AAI,forms a kink and fuses again with CuA; CuA sends off several pectinate branches; CUPprobably does the same; AA and AP are shortly fused basally into A. AA is branched and convex, and AAI fuses with CUP;AP is simple and either concave (primitive),or levelled, or convex (derived). Jugal veins are lost and the jugal area is almost completely reduced. The odonatoid basic apomorphic character is the veinal anal brace fused with the kink in CUPand this in turn with CuA, so that resulting convex vcin is composed of portions of A, AA, AA' , AAI, CUP. CuA, and sends off a (pectinate) series of CuA, CUP,and AA branches, which cannot be distinguished from each other. Note many additional kinks, which in odonate venation followed the kink in CUP,and which were built up through thc accumulation of small evolutionary changes (autapomorphies). Odonata: Gomphidae: Au.strogomphu.s sp.. Recent. Australia. Original drawings by the authors. FIG. 12B. Basal transverse section through the costal triangle of recent dragonflies showing the original vcins PC and CA+, CP-, and ScA+, which became sclerotized next to each other, through autapomorphy. After Kukalova-Peck 1983. FIG. 12C. Dorsal and ventral view of nodal region in recent dragonflies showing differences in veinal patterns. The V-shaped indentation of CP- forming reinforcement of the nodal break and then fusing into PC and C, as well as the kink in ScP supporting distally the nodal break (N), arc visible only ventrally. Anisoptera: Petaluridae: Uropetala carovei (White). New Zealand. Original drawing by Kukalovi-Peck. CAN. J. ZOOL. VOL. 62. 1084 RlEK AND KUKALOVA-PECK

costal triangle .-..MA(L~~+Sc

12A

( costal margin)

dorsal N

ventral Y\ CAN. J. ZOOL. VOL. 62. I984

;zV! s * E"G - 0 gs PE 2 GF 6) .s & s 3% g'u 0 2 .s 5 .. g 242= 6)- C) - 3.@.? .z LL 3 sia, $WC .s g a,. . F2 24. I z.9 2 cL4 L-3. v;.C 5 2 = 12 g RlEK AND KUKALOVA-PECK

FIGS. 17 - 19. Meganisoptera: Meganeuromorpha: Eugeropteridae Riek n. fam.: C;eroptcron crrc-unturn Kiek n. gen. et sp. Holotype, fragment of hind wing, ?Lower Namurian of Argentina. Fig. 17. Hind wing. length of fragment 45 nim. width of fragment 12 mm. Fig. 18. Hind wing in different lighting. Fig. 19. Enlarged hind wing base. Photographed by E. F. Riek. also be found starting from a medial basivenale as two indepen- "wall" of braces are lluted so that their individual position is dent sectors, + and -, possibly in some Namurian or older built into a tluted pattern of a higher rank, thus forming the fossils. Fluting of RP has the capacity to change from concave special tluting of the wing base (Kukalova-Peck 1983). to levelled and to convex by thickening of the dorsal wing Wing venation, articulation, and the alar musculature of cuticle when there is an aerodynamic need. because the RP odonatoids diverged from the pterygote ground plan more than stem is known to have changed from concave to levelled in in any other order, and the chain of transformations had already fossil and recent Ephemeroptera (Fig. 10) and in some Paleo- started in the Early Carboniferous. The change in convexity of dictyoptera (Fig. 6). Other than that, RP branches in Neoptera RA is seen here as one of the many autapomorphies. It occurred are known to have changed from a mildly concave position in in Westphalian A in Erc~siptc~ron,together with fusions be- very primitive Paleozoic Neoptera, to a levelled or even convex tween RP and MA and M and CuA, which replaced arches and position in many recent Neoptera, etc. Other veinal stems, M struts. These innovations gave Erusipteron the odonatoid ap- and Cu, are also known to change fluting from convex to pearance in spite of the fact that other veinal characters re- concave at subordinal levels; veinal stems proximally from the mained very primitive. Change of fluting of RP in advanced FIG. 20. Generalized ephemeropteroid wings were extremely similar to generalized eugeropterid odonatoids (Figs. 1-3) in that the foci of reinforcement MA and RP, and CUAand M were in both marked by veinal arches and struts, which preceded veinal fusions; the shared apomorphy was the veinal anal brace AA and AA' and AA, fused with CUP in an area important for flight; in mayflies this was probably expressed as a bulla. Note that the ephemeroid intercalar veins are already formed, but crossveins are still very irregular, with residual archedictyon as in early odonatoids, and the presence of large hind wings. This giant,oldest known had a wing span of ca. 42 cm and was found in the same thin layer of volcanic ash as a giant, young mayfly nymph 10 cm long (measured without caudal filament and cerci). Ephemeroptera, undescribed family, from a paper in preparation by J. Kukalova- Peck. Middle Upper Carboniferous (Westphalian C) of Central Bohemia. Ventral view. Original drawing by J. Kukalova-Peck. FIG. 21. Tracheae in the wings of recent dragonflies simulate the venational pattern of Paleozoic odonatoids. Note the medial trachea m starting in the middle between r and cu tracheae and then diverging posteriorly towards Cu. The anal and cubital tracheae are close basally. Odonata: Tramea abdominalis, hind wing, dorsal view. TA, tracheal arch; air sacks hatchured. After Brodskyi I981 with lettering changed. FIG.22. Trachea in the wings of recent cockroaches show a very similar start of sc, r, m, cu, and a tracheae from the tracheal arch as in Odonata, but the medial trachea m diverges anteriorly towards R as typical for Neoptera. : Panchlora antillarum, hind wing, dorsal view. After Brodskyi 198 1 with lettering changed. odonatoids was probably useful in strengthening the basal crossing"); CUP becomes shortly fused with A and forms a stems of the radial axial vein, which in odonates is fused to the zig-zag kink; anteriorly, CUPfuses shortly with CuA; the anal posterior axillary plate (other veins are mostly hinged) and brace is composed of AA starting from the axillary plate and is responsible for effortless gliding (Kukalova-Peck 1983). AA, fused to a kink in CUP,which in turn is fused to CuA, and Modern dragonflies often alternate quick wingbeats with all veinal sections (AA, AA,, CUP, and CuA) are aligned so gliding intervals, especially while hunting over large expanses that they form a single, composite, convex vein; branches of of water. this composite vein are often pectinate, alternating with inter- calar veins, so ,that original branches of CuA, CUP, and AA Corrected interpretation of odonatoid venation cannot be distinguished from each other any more (Fig. 12). According to clues found in the wing venation of the oldest, most generalized odonatoids of Namurian, Westphalian, and Tracheation of adult Odonata Lower Permian age, the veinal pattern of modem dragonflies is The emended interpretation of odonatoid venation offered derived from the same ground plan as in other pterygotes, all here finds support in the pattern of tracheation in the adults of veins are fully homologous to other pterygote orders, and living Odonata (Fig. 21). While tracheae are no longer consid- none of them is missing. The presently used Tillyard-Fraser ered to be predecessors to veins, they are often helpful in (1938- 1940) venational system, which interprets two veins as detecting the original veinal pattern, because they tend to be absent, is incorrect. conservative in their course (Whitten 1961). While tracheae of The generalized venation of Eugeropteridae, Erasipteridae, juveniles are frequently variable in subsequent instars, trache- and Meganeuridae provides valuable clues for the inter- ation of adults is more stable and meaningful for wing venation pretation of venation in modern Odonata (Fig. 12), as follows. (Brodskii 198 1). In the tracheal pattern of both Paleoptera and The costal triangle is formed by veins composing a costal Neoptera, the cubital and anal tracheae tend to start close to margin (PC, CA, CP, and ScA), which are fused basally (Fig. each other, while the M trachea starts apart, at about midway 12B); on the underside of the wing CP forms a V-shaped of R (Figs. 2 1, 22). In modern Odonata, the M trachea follows indentation forming a ventral reinforcement of the nodal break a similar path as the vein M in ancestral Odonatoidea, Paleo- before it fuses with PC and CA into PC and C, and Sc forms zoic Ephemeroptera, some Paleodictyoptera, and Permothem- a kink under the nodal break before it joins the costal margin istida and approaches CuA or Cu (Figs. 6- 11). It does not (Fig. 12C); sectoral stems RA and RP are adjacent to each other approach R, and veinal fusion of R and M basally does not basally; sectoral stem of RP acquired secondarily a convex occur in Paleoptera. In Neoptera, the M trachea is directed position; M is fused basally completely with Cu and with a obliquely anteriorly and approaches R (Fig. 22). In contrast to short section of CuA; MA is arched anteriorly and shortly the Paleoptera, R and M veinal stems in all known Neoptera are fused with RP (MA crossing is not discernible); CUPand CuA either very close, or adjacent, or fused at the immediate wing diverge posteriorly towards AA (CuA crossing is not discern- base. ible; CUPcrossing is sometimes preserved as a so-called "anal As shown in Fig. 21, the double radius of Odonata is pene- 1 I64 CAN. J. ZOOL. VOL. 62. 19x4 trated by two tracheal stems. This fact supports a conclusion to CuA+, with a built-in bulla in CUP(Ephemeropteroidea) or that the radius was originally a veinal pair, as were the other a kink in CUP (Odonatoidea). The orders of the Paleo- wing veins, and that the veins in the proto-wing diverged dictyopteroidea share instead a different basic apomorphic dichotomously from each other immediately at their start from character: a sclerotization in the basal corner of anal area, the basivenale (Kukalova-Peck 1983). The double radius, to which is sometimes delimited by an oblique or a V-shaped bar. our knowledge, never occurs in Neoptera, but it is present as Anal branches crossing the sclerotization vary in number (Figs. a plesiomorphy in Odonatoidea, Ephemeroptera, and in some 6, 7) and sometimes are not discernible. Paleodictyoptera. Since all these orders also became special- The autapomorphic evolution towards a distinctly odonatoid ized very early in their evolution for effortless gliding, it is very venation, only indicated in (?Lower) Namurian, gathered probable that the double radius was advantageous for some momentum, and in Lower Westphalian fossils the "odonatoid types of flight involving gliding, and that this quality "saved" type" of wing was almost formed (Fig. 4). Basic apomorphic this primitive character from complete disappearance from character changes typical for Odonata, such as the shift toward pterygote wings. direct flight musculature or fusion of sclerites into two axillary plates, probably somewhat preceded the changes in venation; Evolution of odonatoid venation Namurian, Westphalian, Stephanian, and Lower Permian Venation of Paleozoic odonatoids is similar to that of dragonflies already have a typically odonatoid articulation with other paleopterous orders of the same era. Veinal characters anterior and posterior axillary plates (Figs. 1, 5, 11). While such as separated radial stems and similar pattern of short the prey was increasing in size during the Westphalian and veinal fusions (M, or MA and RP. M and CuA, or M and Stephanian, so were the odonatoid predators. Cu, and AA and CUP) are present or indicated also in some Modern odonatoid venational pattern is dominated by Paleodictyoptera, Megasecoptera, Permothemistida, Diaphan- "kinks." While generalized Carboniferous dragonflies had only opterodea, and Ephemeropteroidea (Figs. 6- 10). one zig-zag kink in CUP, Mesozoic and Cenozoic dragonflies In contrast to this, the pattern of fusions and braces in Neo- also developed a kink in ScP (at nodus), a kink in MA and RP ptera is distinctly different. The most important features of (at arculus), a kink in MP (at triangle), and a kink in CuA neopterous venation are the stems of R and M adjacent or fused (beyond subtriangle) (Fig. 12). Succession in development of near the wing base representing, in all probability, one of the kinks during the Mesozoic should be of great help to taxon- basic apomorphic characters of Neoptera. The neopteroid anal omists studying the Recent epoch, but this is beyond the realm brace is veinal, composed of sections of the AA vein as in of this paper. ephemero-odonatoid group, but its fusion with CUP is incon- As proposed by several previous authors and documented spicuous and not important for flight as in the latter (Kukalova- by various morphological characters (Sharov 1966; Kukalova- Peck 1983, Figs. 17A, 17B). In fact, the last leg of AA is weak Peck 1978, 1983; Hennig 198 1) Odonatoidea and Ephem- and sometimes crossed by a claval fold. However, the neop- eropteroidea are probably sister groups; the ephemero- terous anal brace developed a new and important function: to odonatoid stem group and the paleodictyopteroid stem-group protect the anal area when the wings are folded in the resting are also believed to be sister groups. This conclusion is sup- position on the back. For that purpose, it often acquired addi- ported by the present study of odonatoid venation. It also leads tional sclerotization. Again, this adaptation of the anal brace is to the conclusion that gliding in Paleoptera (and all other Pter- unknown in Paleoptera, even if they fold their wings com- ygota) is a grade and evolved twice, in ephemero-odonatoids pletely backwards as do some Diaphanopterodea (Kukalova- and gliding paleodictyopteroids. Peck 1983). The neopterous pattern of veinal fusions is rich and From the findings documented in this paper and summarized varied, in fact, much more varied than that of the Paleoptera. above, it is concluded thal (i) insect wings are monophyletic, In addition to characteristic fusion of R and M basally, there is (ii) Paleoptera is a valid phylogenetic unit and not merely a usually present an M and Cu brace executed by a crossvein, a grade, (iii) Odonatoidea are to be referred to Paleoptera, bar, or by a veinal fusion of some sort. However, there are also (iv)Odonatoidea and Ephemeropteroidea are sister groups, and primitive neopterous wings which completely lack this charac- (v) gliding in Paleoptera, as well as in Neoptera, is a grade. ter (i.e., some Paleozoic Liomopteridae of the plecopteroid The Till yard- Fraser interpretation of odonatoid venation stem group). In Paleoptera, fusion between the stems R and M led, unfortunately but inevitably, to the erroneous conclusion at the base is absent from the wings of gliding Paleoptera, of some authors that odonatoid venation and tracheation is Paleodictyoptera, Permothemistida, primitive Megasecoptera, similar neither to that of the Paleoptera nor the Neoptera, and Ephemeroptera, and Odonata (Figs. 6- 1 1). R and M stems in that insect wings are thus polyphyletic. However, from the Paleoptera became adjacent basally only in limited circum- point of view of general evolutionary evidence, this idea is no stances, namely if the veins are crowded in the petiolate wings more acceptable than a polyphyletic origin of tetrapod verte- of specialized Megasecoptera, or in the wings of some brate limbs. Insect wings and limbs of vertebrates have two Diaphanopterodea, which are small and fold backwards at rest. features in common: their morphology is fully homologous in In comparison with primitive Paleodictyoptera, venation of all taxa and the similarity of their morphology increases in the the oldest presently known odonatoids is not as rich in branches fossil ancestors of living groups. It should always be kept in and is comparable to more advanced Paleodictyoptera. There is mind that fossil evidence records actual states of characters, as little doubt that gigantism of Late Paleozoic Meganisoptera is they passed through successive previous stages. Therefore, the secondary, because the wings are "filled out" by numerous gradual increase of overall similarity in older and older ances- secondary branches and crossveins, which formed from the tors provides direct evidence of their monophyletic origin. In membrane (Fig. 5). spite of its defects and the many artifacts of preservation, the In addition to the plesiomorphic-paleopterous veinal pat- fossil record provides a unique factual basis against which an tern, Odonatoidea and Ephemeropteroidea share an apo- evolutionist can test interpretations and from which he or she morphic veinal anal brace AA reaching from the axillary plate can improve an understanding of evolution. RIEK AND KUKALOVA-PECK 1165

Acknowledgements Harv. Univ. 108: 777-827 The unique specimens of the most primitive dragonflies were 1960. Studies on North Amcrican Carboniferous insects. I. collected by paleobotanists Dr. S. Archangelsky and Dr. R. The Protodonata. Psyche. 67: 98- 109. COMSTOCK,J. H., and J. G. NEEDHAM.1898. 'The wings of insects. Leguizamon, of the Universidad Nacional de La Plata, Facul- Am. Nat. 32: 43-48, 81-89. 231-257, 335-340, 413-424, dad de Ciencias Naturales y Museo, Paseo del Bosque, La 561 -565. 769-777, 903-91 I. Plata, Argentina. These finds were recognized as important and 1899. The wings of insects. Am. Nat. 33: 117-126. were brought to the attention of one of us (E.F.R.) by ento- 573-582, 845-860. mologist Dr. R. A. Ronderos of the Universidad Nacional FORBES,W. T. M. 1943. The origin of wings and venational types of de La Plata. We would like to express our gratitude to these insccts. Am. Midl. Nat. 29: 381 -405. Argentinian scientists, whose concerted efforts provided, at FRASER,F. C. 1938. A note on the fallaciousness of the theory of last, the clues to the long unsolved problem, the relationship of pretracheation in the venation of Odonata. Proc. R. Entomol. Soc. the wing venation of dragonflies. We are also indebted to Dr. London. Ser. A, 13: 60-70. L. N. Pritykina, Paleontological Institute of the Academy of 1948. A new interpretation of the course of the subcostal vein in the wings of Odonata, with remarks on Zalessky's notation. Proc. Sciences USSR, Moscow, for sharing with one of us (J.K.P.) R. Entomol. Soc. London, Ser. A, 23: 44-50. her unpublished results on the meganisopterid pleuron, and to 1957. A reclassification of the order Odonata. Royal Zoo- Dr. 0. S. Flint, Smithsonian Institution, Washington, who rec- logical Society, Sydney, N.S. W. pp. 1 - 133. ommended that we study the material. Mrs. E. J. Lambton of HAMILTON,K. G. A. 197 1. The insect wing. Part I. J. Kans. Entomol. Carleton University is thanked for typing the manuscript. This SOC.44: 421-433. work was partly supported by operating grants from the Natural 1972. The insect wing. Parts 11, Ill, and IV. J. Kans. Ento- Sciences and Engineering Research Council of Canada to mol. Soc. 45: 54-58. 145- 162, 295-308. J.K.P. The Department of Geology at Carleton University pro- HENNIG.H. 198 1. Insect phylogeny. John Wiley & Sons, New York. vided research space and facilities as well as encouragement. pp. 1-514. KRISTENSEN.N. P. 1981. Phylogeny of insect orders. Annu. Rev. ARCHANGELSKY,S., and R. R. LEGUIZAMON.197 1. Vojno~lskyct Entomol. 26: 135-157. ctrgentinct n. sp., a new gymnosperm from the Upper Carboniferous KUKALOVA,J. 1964. To the morphology of the oldest known of Sierra de 10s Llanos. La Ria-ja province. Ameghiniana, 8(2): dragonfly Erctsipteron ltrischi Pruvost, 1933. Vestn. Ustred. 65-72. Ustavu Geol. 39: 463-464. AZCUY,C. L. 1975. Namurian land Westphalian miospores from 1969. Revisional study of the order in the Malanzan - Loma Larga region, La Ria-ja province, Argentina. I. Upper Carboniferous shales of Commentry. France. Part I. Psyche, Geographic and geologic position of the region and systematic 76(2): 163-215. description. Ameghiniana. 12( I): 1 -69. KUKALOVA-PECK,J. 1970a. Revisional study of the order Palaeo- BOUDREAUX,H. B. 1979. phylogeny, with special refer- dictyoptera in the Upper Carboniferous shales of Commentry, ence to insects. John Wiley & Sons, New York. pp. 1 -320. France. Part 11. Psyche. 76(4): 439-486. BRAUCKMANN,C. 1983. Ein lnsektenrest (Odonata, Meganisoptera) 1970b. Revisional study of the order Palaeodictyoptera in aus dem Ober-Karbon des Piesbergcs bei Osnabriick. Osnabriicker the Upper Carboniferous shales of Commentry, France. Part 111. Naturwiss. Mitt. 10: 7- 14. Psyche, 77( I): 1-44. BRODSKII,A. K. 1970. Organization of the flight system of Ephemera 1974. Pteralia of the Paleozoic insect orders Palaeo- vulgatct L. (Ephemeroptera). (In Russian.) Entomol. Obozr. 49: dictyoptera, Megasecoptera. and Diaphanopterodea (Paleoptera). 307-315. Psyche, 81(3-4): 416-430. BURNHAM,L. 1983. Studies on Upper Carboniferous insects: I. The 1975. Megasecoptera from the Lower Permian of Moravia. Geraridae (Order Protorthoptera). Psyche. 90( 1 - 2): 1 - 57. Psyche. 82(1): 1-19. CARLE,F. L. 1982. The wing vein homologies and phylogeny of 1978. Origin and evolution of insect wings and their relation the Odonata: a continuing debate. Soc. Int. Odonatol. (Rapid to metamorphosis, as documented by the fossil record. J. Morphol. Commun.), 41: 1-66. 15: 53 - 126. CARPENTER,F. M. 193 !. The lower Permian insects of Kansas. Part 1983. Origin of the insect wing and wing articulation from the 2. The orders Palaeodictyoptera, Protodonata, and Odonata. Am. J. arthropodan leg. Can. J. Zool. 61: 1618-1669. Sci. 21(5): 97- 139. KUKALOVA-PECK,J., and S. B. PECK. 1976. Adult and immature 1938. Two Carboniferous insects from the vicinity of Mazon Calvertiellidae (lnsecta: Palaeodictyoptera) from the Upper Paleo- Creek, Illinois. Am. J. Sci. 36: 445-452. zoic of New Mexico and Czechoslovakia. Psyche, 83(1): 79-93. 1939. The Lower Permian insects of Kansas. Part 8. KUKALOVA-PECK,J., and E. S. RICHARDSON,JR. 1983. New Homo- Additional Megasecoptera, Protodonata, Odonata, Homoptera, iopteridae (lnsecta: Paleodictyoptera) with wing articulation from Psocoptera, , Plectoptera and Protoperlaria. Proc. the Upper Carboniferous of Mazon Creek, lllinois. Can. J. Zool. Am. Acad. Arts Sci. 73(3): 29-70. 61: 1670- 1687. 1943a. The Lower Permian insects of Kansas. Part 9. The LA GRECA,M. 1980. Origin and evolution of wings and flight in orders , Raphidiodea, Caloneurodea, and Protorthoptera insects. Bull. Zool. 47(Suppl.): 65-82. (Probnisidae), with additional Protodonata and Megasecoptera. LAMEERE,A. 1922. Sur la nervation alaire des insectes. Bull. Acad. Proc. Am. Acad. Arts Sci. 75: 55-84. Sci. R. Belg. 5(5): 138- 149. (English translation in Psyche, 30: 1943b. Studies of Carboniferous insects from Commentry, 123- 132, 1923.) France. Part 1 , introduction and families Protagriidae, Mega- LEMCHE,D. 1940. The origin of winged insects. Vidensk. Medd. neuridae and Campylopteridae. Bull. Geol. Soc. Am. 54: Dan. Naturhist. Foren. 104: 127 - 168. 527 -554. MARTYNOV,A. V. 1924. [Jber zwei Grundtypen der Fliigel bei den 1944. Carboniferous insects from the vicinity of Mazon lnsekten und ihre Evolution. Z. Morphol. Oekol. 4: 465-501. Creek, Illinois. Ill. State Mus. Sci. Pap 3(1): 9-20. MATSUDA,R. 1970. Morphology and evolution of the insect thorax. 1947. Lower Permian insects from Oklahoma. Part 1. Intro- Mem. Entomol. Soc. Can. 76: 1 - 43 1 . duction and the orders Megasecoptera, Protodonata, and Odonata. 1981. The origin of insect wings (Arthropoda: lnsecta). lnt. Proc. Am. Acad. Arts Sci. 76(2): 25-54. J. lnsect Morphol. Embryol. 10(5/6): 387-398. 1954. Classification of insects. Bull. Mus. Comp. Zool., NEEDHAM.J. G. 1903. A genealogic study of dragonfly wing vena- I166 CAN. 1. ZOOL. VOL. 62. 1984 tion. Proc. U.S. Natl. Mus. 26: 703-764. tion of bridges. Proc. Linn. Soc. N.S.W. 41: 871 -887. 1951. Prodrome for a manual of the dragonflies of North 19 17. The biology of dragonflies. Cambridge University America, with extended comments on wing venation systems. Press, Cambridge. pp. 1 -396. Trans. Am. Entomol. Soc. 77: 2 1-62. 1922~.New researches upon the problem of the wing vena- PRITYKINA,L. N. I98 1. New Triassic dragonflies from Middle Asia. tion of Odonata. I. A study of the tracheation of the larval wings in Tr. Paleontol. Inst. Akad. Nauk SSSR, 183: 5-42. the genus Uropetclla from New Zealand. Entomol. News, 33: 1 -7. PRUVOST,P. 1933. Un ancetre des Libellules dans le terrain houiller 1922b. New researches upon the problem of the wing vena- de TchCcoslovaquie. Ann. Soc. Geol. 58: 149- 155. tion of Odonata. l. A study of the tracheation of the larval wings RASNITSYN,A. P. 1980. Origin and evolution of . Tr. in the genus Uropetcllcl from New Zealand. Entomol. News, 33: Paleontol. Inst. Akad. Nauk SSSR, 174: 1 - 190. 45-51. RIEK,E. F. 1970. Fossil history (of the insects). In The insects of 1925a. Kansas Permian insects. Part V. The orders Proto- Australia. Edited by CSIRO, Canberra. Melbourne University donata. Am. J. Sci. 10: 4.1 -73. Press, Melbourne. pp. 1 - 1029. 1925b. The British Liassic dragonflies. Publications of the ROHDENDORF,B. B. 1962. Arthropoda. Insecta. In Fundamentals of British Museum, Fossil Insects, Vol. 1. pp. 1-38. Paleontology. (In Russian.) Academy of Sciences USSR, Moscow. 1926. Insects of Australia and New Zealand. Angus and pp. 29-374. Robertson, Sydney. pp. 1 - 560. ROHDENDORF,B. B., and A. P. RASNITSYN.1980. Historical devel- 1928. The evolution of the order Odonata. Part 1. lntro- opment of the Class Insecta. (In Russian.) Tr. Paleontol. Inst. duction and early history of the order. Rec. Indian Mus. (Calcutta), Akad. Nauk SSSR, 175: 1-256. 30(2): 15 1 - 172. SHAROV,A. G. 1966. Basic athropodan stock. Pergamon Press, New 1935. Upper Permian insects of New South Wales. IV. The York. 1-271. order Odonata. Proc. Linn. Soc. N.S.W. 60: 374-384. SINICHENKOVA,N. D. 1980. Revision of the Order Permothemistida TILLYARD,R. J., and F. C. FRASER.1938- 1940. A reclassification (lnsecta). Paleontol. Zh. 4: 91 - 106. of the order Odonata based on some new interpretations of the TILLYARD,R. J. 1914. On some problems concerning the devel- venation of the dragonfly wing. Aust. Zool. 9: 125- 169, opment of the wing venation of the Odonata. Proc. Linn. Soc. 195-221, 359-390. N.S.W. 39(1): 163-216. WHITTEN,J. M. 1962. Homology and development of insect wing 191 5. On the development of the wing venation in zygop- tracheae. Ann. Entomol. Soc. Am. 55(3): 288-295. terous dragonflies, with special reference to the Calopterygidae. ZALESSKY,G. 1944. A brief statement of reasons for some changes in Proc. Linn. Soc. N.S.W. 40: 2 12-230. the terminology of insect wing-venation. Proc. R. Entomol. Soc. 19 16. Further researches upon the problems of the radial and London, Ser. A, 19: 37-47. zygopteroid sectors in the wings of Odonata, and upon the forma-